diy solar

diy solar

Sol-Ark 12K with AC coupling

Yeah, I got lucky there. Everything is right in that corner.

0000 has acceptable (by code) 2.79% vd over 40', but 80' of that would add up quick!
 
35 feet is getting quite far to run 150 amp battery cables. Would it make sense to also put the inverter/charger in the battery shed and run 120/240 split phase to/from the shed? the DC coupled solar would still need to run out there as well, but AC coupled could stay in the garage. If I did this with my setup, I would need to run eight #6 gauge wires, 4 for the main feed to the my XW and then 4 back to the Essential Loads panel. When it goes off grid, the transformer in the battery inverter is creating the neutral, so it does need to be able to carry all of the current. Might be able to use one ground wire, but I would run both neutrals for sure.
 
I'm going to have to find out what the inspection process is like here. I want to be to code and legal.
But from what I see, a lot of people have a pile of messy looking batteries with wires running all over the place. :confused:
 
Yeah, at 35 feet I would probably run the AC that distance and mount the inverter with the batteries. I was just saying it's possible.

On the neutral. Doesn't the neutral just connect to a bus inside the unit? I thought the neutral was common and wouldn't need two dedicated wires.
 
In theory, if the load is balance on both legs, the neutral carries almost no current. Of course, theory and reality don't always match up. If the load is unbalanced, and the there are some reactive components from things like motors and non PFC power supplies, the neutral current can get up to the full power of either leg. This is why they call for the same size wire on the neutral. Legally, you may get away with just running a single neutral out to the inverter at the batteries. All of the single leg 120 volt loads can tie to the neutral back in the house/garage. When running on grid, the hot side would be going out to the inverter, through the transfer relay, and back to the house, so it would act just like a switch leg. When the power is coming from the inverter, the hot leg is coming from the inverter's internal transformer, and the return current will be on the neutral wire between the inverter and back to the main panel neutral buss bar. An inspector seeing a conduit with 4 hot wires and only one neutral might question that though. Having the second neutral wire tying the inverter to the neutral buss bar in the backup loads panel is certainly not going to hurt anything. I have a neutral from the main panel to the inverter, and from the main panel to my backup loads panel, and then I also tied the inverter output neutral to the backup loads panel.
 
I agree that the neutral is important and needs to be able to carry the full load. When off grid the neutral path for my system would be like you said: XW > main panel > sub panel/el panel. I don't plan to run the neutral from the XW to the sub panel (I'll wait to see what the inspector says). If I run the second neutral to the sub panel it would literally run in the same conduit as the first neutral from the main panel. The XW will mount on the inside wall behind the main panel (which faces outside) and the sub panel is 30 ft away.
 
I am not using a Sol-Ark, but I am now looking at doing a split setup. Adding some DC coupled solar to my AC coupled system.

There are several advantages to having some of both.

AC coupled micro inverters are currently the most efficient at taking solar and putting AC into your home electrical system while the sun is shining, but if you want to store some power to use at night, it has to be converted back to DC to charge the batteries, and then converted again back to AC to feed your loads. In my current AC coupled only system, I am losing about 1 KWH every day from the loss of charging and discharging the battery due to all of the conversions.

So here is my thought. Have the micro inverters supply the power you need during the day, use it as it is produced at maximum efficiency, and maybe sell some to the grid if you over produce and have net metering. Then have DC solar panels to charge the battery during the day. A charge controller is more efficient as it only have to convert the voltage to match the battery bank. Then at night, you run off the battery power and only do one conversion to AC.

The other huge advantage is the system can now "dark start". If there is a power failure, my AC coupled setup does work just fine. Any extra power from the AC microinverters does charge the battery bank while the sun is up. Once the sun goes down, I run off the batteries. But what happens when the battery bank runs down over night? The inverter has to shut down. Then when the sun comes up, there is no "grid" for the solar microinverters to sync up to. The system does not come on without some human intervention. With DC coupled solar panels, they will start charging the battery bank as soon as sun starts hitting the panels. When the batteries charge up enough, the inverter will start back up. Then you will have your local "grid" and the AC coupled inverters will then be able to qualify the power and turn on again.

I did look into a Sol-Ark, but from the install documents I looked at, it did not look like it would even work properly with only AC coupling. They always recommend having some DC coupled solar feeding into the Sol-Ark.
 
My system also does not have an issue if the grid is up, the dark start issue is only during a grid outage. I have mind set to shut down with a solid 30% still left in the battery bank, but when the sun comes up, I would still need to log in from a battery powered laptop and change the shut down setting lower to wake it up and get the AC coupling to work again. I would expect the Sol-Ark with only AC coupling would have the same issue.

The charge side on my Schneider is rated at 92% efficient, so 8% loss charging from AC, not too bad, but this is also after the 96% efficient Microinverters, so another 4% loss there. A good MPPT charge controller can be 98% efficient, just 2% loss. All together, 7% to 10% less loss than DC to AC to DC again. When you push 10 KWH to the battery, that is nearly a full KWH lost. For my system, I am looking at adding nine 300 watt panels, wired 3S3P. The current will peak a little under 30 amps and 90 volts. I will run just 25 feet of #8 after the combiner, so loss will not be a problem. I looked at a higher voltage charge controller, but could not justify the cost difference. Putting all 9 panels in series would have been just 10 amps at 270 volts, but that is right at the bottom of the MPPT range on the 600 volt controller. The 150 controller is 1/3 the price.

I know the Sol-Ark has a built in MPPT, mine does not.

I can see some reasons why you might want to hide that you have solar, but that was never an option for me. My original grid tie Enphase system went in without storage, and not doing net metering was just not an option. I would have had to curtail 50% of my production during the day and just lose it. Before the battery system went in, I was exporting over 10 KWH every day, and buying it back at night. Now with the battery, I charge over 8 KWH into the battery and run off of that during the peak time of use rate. I still end up exporting some during the day. Once the battery is full, and the solar panels are still making more power than the house needs, you have 2 choices. You can curtail the output and get nothing for the extra capacity, but you then don't need net metering, and you may be able to have a cheaper rate when you do need to buy some power. OR.. You can export that extra power and get some credit for it when you need to buy some power. With my current 16 x 300 watt panels, I have managed 2 months where my electric bill was less than zero and had credit carry to the next month. Most months, I still end up buying $50 to $150 worth of power when I have to run my air conditioning. We have had over 2 weeks so far where it has been over 100F each day. With the 9 more panels to make power for use at night, I should zero my bill much more often. 2,700 watts added to my existing 4,800 watts will get me very close to grid zero on all but the hottest days. When it is crazy hot, I also lose a bit of solar production, with my peak power falling off nearly 400 watts.
 
The all in one inverters, like Sol-Ark, still rely on batteries to filter the AC current profile load on PV charge controller so you have the downside of more stress on batteries due to ripple current.

Someone will probabably chime in with some of these hybrid inverters allow battery-less operation. This is more a marketing gimmick then a useable function. They sort of work if you have massive amount of excess PV power available. Not a practical option for grid down backup power.
 
There are a few systems that sort of work without a battery, but it really does not make sense. For them to work, they do need at least a decent capacitor bank to allow it to stabilize the power, and it needs to throw away a lot of excess potential solar generation. With even a small battery, you can store some of the extra solar, and have it meet surges, and also keep running when a cloud passes. Not having a battery just handicaps the whole system.
 
The all in one inverters, like Sol-Ark, still rely on batteries to filter the AC current profile load on PV charge controller so you have the downside of more stress on batteries due to ripple current.
Can you quantify that stress? My only reference to ripple current was the difference between an Outback Radian and an Outback Skybox three years later. The ripple current from the Radian was so bad that the CT on my Orion Jr. BMS could not provide a meaninful SOC but the same BMS on the Skybox was accurate. I realize this is anectdotal but sufficient for me to conclude that the Ripple current from the Skybox was negligable.
 
For single phase AC sinewave inverters, the ripple current on the inverter DC input goes from near zero (inverter idle overhead) to a peak of about 2.3 times the average of DC current. It is at twice the AC inverter frequency.

The PV charge controller produces a constant DC output proportional to illumination level on panels. The instantaneous difference from the illumination based constant current the PV puts out and the sinewave inverter input current profile is made up by attached batteries, like a large capacitor. For high wattage inverter AC output to grid push, high peak amperage is pumped in and out of battery. Net average current to battery is near zero but the high amperage ripple in-out current still puts stress on batteries.

The battery AH must be large enough to handle the peak ripple current even though the average battery current is near zero.
 
That is not exactly the quantified answer I was looking for. In the meantime, I will have to rely on my anectdotal information. In the absence of anything quantifiable I won't worry too much about the stress on my batteries. It beats the alternative of having a $300 a month electrical bill. Also, my 42 kWhr pack helps me reduce that bill by load shifting..
 
Last edited:
My 6kW Sunny Islands are supposed to have at least 100 Ah 48V lead-acid or 50 Ah lithium per inverter. (With 405 Ah and 4 inverters, that's all I have.) That indicates what they want for surge and ripple.

I took current readings at 40% load (decided it was too much bother to recommission with half turned off for 80%) using 10kW duct heater. Ripple was a significant percentage of DC average current. I estimate that at 100% load, battery current will look exactly like rectified AC it is producing (scaled for voltage). Current will drop to about zero from battery during AC zero crossings. The capacitors can't keep up with that, only with the higher frequency PWM used to synthesize sine wave.

3-phase would have its advantages.

Of course, in my system that ripple would only happen at night, when battery supplies power. During the daytime, 100% of AC comes from PV inverters. Unlike the low-frequency battery inverters with capacitors at battery voltage, GT PV inverters have capacitors at MPPT voltage. They deliver sine wave current drawing from capacitors, which can release energy by dropping in voltage. No low internal resistance battery to hold them up.

It is an interesting question - for a PV system with SCC, battery, and inverter, if battery is charging and discharging at about 1.2C 120 times per second, what does that do to life?

I think a high-frequency inverter may or may not have such ripple on the battery. It could boost a steady current into a high voltage capacitor, and let that capacitor ripple as it is drawn from to deliver sine wave. Just like my PV inverters. Alternate design could vary the current boosted from battery 120 times per second to hold constant high voltage rail. Yet another would just high frequency boost from battery to sine wave.

The 400V battery designs (PowerWall, Sunny Boy Storage) would draw ripple current from battery.
 
Can you quantify that stress? My only reference to ripple current was the difference between an Outback Radian and an Outback Skybox three years later. The ripple current from the Radian was so bad that the CT on my Orion Jr. BMS could not provide a meaninful SOC but the same BMS on the Skybox was accurate. I realize this is anectdotal but sufficient for me to conclude that the Ripple current from the Skybox was negligable.
How big was your battery at the time....just asking for a reference...I will check the AC on my 20kwh battery on my Radian....
 
I certainly do see a fair amount of ripple current with my Schneider XW-Pro. The current reading on the BMS bounces about 2-3 amps when it is charging or discharging around 30 amps. When I checked it with an AC ammeter, it actually had nearly as much 120 Hz AC current as the steady DC current. I added a 10,000 uF capacitor in the batter housing on the output side of the BMS, and that did reduce the amount that the current reading fluctuates. The ripple current on the battery cables to the inverter did not change, but between the BMS and the battery cells it dropped it a little, maybe 3-5% less. So maybe adding a full Farad might make a real difference.
 
Current bouncing around is usually due to commonly occuring small fluxuations of grid voltage and inverter's reaction time to the variance. Hybrid inverters are bi-directional and have to continuously adjust their PWM to maintain grid voltage and phase lock which has a little feedback loop delay in reacting to input AC changes (grid or AC generator) changes. The correction delay results in momentary current push or pull from the battery. Voltage feedback bandwidth is faster than very slow feedback bandwidth for phasing. This is why unstable warbling generator engine speed causes hybrid inverter to disconnect.

The current read by battery monitors shunt voltage output is highly averaged to filter out the 120 Hz ripple current, leaving just the average current.

10,000 uF won't do much of anything when dealing with 50-100 amps of battery line ripple current at 120 Hz. To ride across 8.33 millisecs with 50 amps ripple current will take like 0.2 Farads. (200,000 uF)
 
Ripple depends on how stiff the battery is, and how much wire resistance.

If the battery had zero ohms internal resistance and the wiring too, voltage at inverter and across capacitors wouldn't vary at all, so zero current supplied by capacitors. 100% of ripple current comes from battery.

If battery and wires have a lot of resistance, higher ripple voltage and more of the ripple current drawn from inverter comes from capacitor rather than battery.

My theory had been that lower IR lithium cells would end up carrying more of the ripple current, because they are about 1/5th the resistance of lead-acid. But it looks to me like at 100% load of inverter, about 100% of ripple is supplied even by lead-acid. At some lower load, lithium would deliver high ripple current before lead-acid.

10,000 uF won't do much of anything when dealing with 50-100 amps of battery line ripple current at 120 Hz. To ride across 8.33 millisecs with 50 amps ripple current will take like 0.2 Farads. (200,000 uF)

With about 2V ripple (1F = 1 amp second per volt). If you draw inverter end of battery cable down 2V, how much current will be drawn from battery? 2V/50A = 40 milliohms. Compare to battery IR and cable resistance. I think lithium battery (cell spec 0.25, typical 0.17 milliohm) would be 4 milliohms for 16s.

Seems to me caps on input of battery inverter are mostly for the higher switching frequency.
 
At the cells, there is virtually no voltage ripple as the internal resistance of these NMC cells are incredibly low. I have about 8 feet x2 of 2/0 cable between the inverter terminals and the battery bank. On a typical day, my battery charge current is only 25 amps, and my maximum discharge current hits just over 80 amps.

Using my true RMS Fluke clamp ammeter, it should a very steady current during charging, and when the load is steady as well. My JK BMS seems to be sampling the current and averaging a few readings, but it is not sync'd to the 60 hz line frequency, so the samples are all over different parts of the 120 hz half wav current waveform. The multiple sample averaging does filter it out pretty good, just leaving the 2-3 amp fluctuation in the reading. It goes above and below when my Fluke reads, so the total over time seems quite good.

I agree, the 10,000 uF is not enough to "fix" the issue, but I did see a small change. I don't remember the number, but I did see a fair bit of ripple current going into the capacitor, so it is trying to help. The huge 10,000 can I have was the only large cap in my collection that is rated over 60 volts. It is actually 150 volt. It is 80 mm in diameter and 300 mm tall. The cap is on the battery end of the 8 foot cables, so it is an RC filter to the battery bank. I figured a full 1F cap would do a lot more, but they get expensive to stack up for 60 volts.
 
Back
Top